Linear and Nonlinear Structure-Retention Relationship Analysis of Different Classes of Pesticides Isolated From Groundwater

Abstract

The present article gives an insight into the linear and nonlinear relationships between the retention behavior in reversed-phase high performance liquid chromatography (RP-HPLC) of several classes of pesticides isolated from groundwater, and their in silico physicochemical, topological and lipophilicity molecular descriptors. The quantitative structure-retention relationship (QSRR) chemometric approach was applied for this purpose on a large set of compounds (77 pesticides). The selection of the most appropriate molecular descriptors was achieved by stepwise selection procedure coupled with partial least squares method and the variance inflation in projection parameter (SS-PLS-VIP). QSRR included the linear regression (LR), multiple linear regession (MLR), and artificial neural networks regression (ANN-R). In order to select the optimal QSRR model, statistical validation parameters were used. Additionally, a relatively new chemometric method called sum of ranking differences (SRD) was applied in order to select the optimal regression model. The obtained results showed that certain models can successfully be used for precise prediction of the retention time of the studied compounds.

Keywords:

• artificial neural networks
• chemometrics
• groundwater
• high performance liquid chromatography
• pesticides
• sum of ranking differences

Supplemental Material

Supplemental data for this article can be accessed on the publisher’s website.